def mexico_all_links(driver):
#the mexican website provides advisory in spanish
#we can display the link
url = 'https://guiadelviajero.sre.gob.mx/'
driver.get(url)
soup = BeautifulSoup(driver.page_source, 'lxml')
reg = regex.compile(r'\/103-ficha-de-paises\/')
a = soup.findAll('a', attrs={'href': reg})
links = {}
iso_es = get_iso_es()
LOGGER.info(
f'Retrieving the URLs for all countries for the Mexico advisory')
for att in a:
try:
name = att.text.strip()
iso = iso_es[name]
href = 'https://guiadelviajero.sre.gob.mx' + att['href']
href = '<a href =\'' + href + '\'>Mexican Government Webesite</a>'
links[iso] = {
'advisory_text': href,
'country_iso': iso,
'name': name
}
Logger.success(f'The URL for {name} was successfully retrieved')
LOGGER.success(
'Successfully retrieved the URLs for all countries of the Mexican advisory'
)
except Exception as error_msg:
LOGGER.warning(
f"This country's iso was not found for {name} because of the following error: {error_msg}"
)
#get the visa for mexico like for other countries from wikipedia
LOGGER.info(
'Parsing visa requirements for all countries for the Mexican advisory')
try:
wiki_visa_ob_MX = wiki_visa_parser(wiki_visa_url_MX, driver)
visas = wiki_visa_ob_MX.visa_parser_table()
visas = replace_key_by_iso(visas)
LOGGER.success(
'Successfully parsed all countries for the Mexican advisory')
except Exception as error_msg:
LOGGER.error(
f'Was not successful in parsing visa requirements for Mexican advisory because of the following error: {error_msg}'
)
data = {}
for key in visas:
try:
data[key] = links[key]
info = data[key]
info['visa-info'] = visas[key].get('visa-info')
except Exception as error_msg:
LOGGER.warning(
f'The following iso was not found: {key} because of the following error: {error_msg}'
)
return links
# Pick specific dataframe that will always be the third index
data_table_social = data_tables[3]
LOGGER.info(f'Parsing returned following dataframe: {data_table_social}')
# Get latest year
latest_year_social = data_table_social.columns[-1]
LOGGER.info(f'Parsing returned following year: {latest_year_social}')
# Get data of latest year
social_data = data_table_social.iloc[:,-1]
LOGGER.info(f'Parsing returned following year data: {social_data}')
lifeExpectancy_object = social_data[social_data.index.str.startswith('Life expectancy')]
lifeExpectancy = next(iter(lifeExpectancy_object), 'no match')
LOGGER.success(f'Following currency data was retrieved: {lifeExpectancy}')
infantMortality_object = social_data[social_data.index.str.startswith('Infant mortality')]
infantMortality = next(iter(infantMortality_object), 'no match')
LOGGER.success(f'Following currency data was retrieved: {infantMortality}')
nbOfPhysicians_object = social_data[social_data.index.str.startswith('Health: Physicians')]
nbOfPhysicians = next(iter(nbOfPhysicians_object), 'no match')
LOGGER.success(f'Following currency data was retrieved: {nbOfPhysicians}')
homicideRate_object = social_data[social_data.index.str.startswith('Intentional homicide rate')]
homicideRate = next(iter(homicideRate_object), 'no match')
LOGGER.success(f'Following currency data was retrieved: {homicideRate}')
# Pick specific dataframe that will always be the fourth index
data_table_env = data_tables[4]
# Create table if it does not exist
DB.add_table('emergency', country='text', police='text', ambulance='text', fire='text')
data_tables = pd.read_html('http://chartsbin.com/view/1983')
data_table = data_tables[0]
latest_year = data_table.columns[1]
for country in iso_list_2:
try:
LOGGER.info(f'Getting emergency contacts data for {country}')
if str(data_table.iloc[iso_list_2.index(country)][1]) == 'nan':
police = ''
else:
police = data_table.iloc[iso_list_2.index(country)][1]
if str(data_table.iloc[iso_list_2.index(country)][2]) == 'nan':
ambulance = ''
else:
ambulance = data_table.iloc[iso_list_2.index(country)][2]
if str(data_table.iloc[iso_list_2.index(country)][3]) == 'nan':
fire = ''
else:
fire = data_table.iloc[iso_list_2.index(country)][3]
LOGGER.success(f'Following emergency contacts data was retrieved: {country}: {police} {ambulance} {fire}')
LOGGER.info('Inserting data into database.')
DB.insert_or_update('emergency', country, police, ambulance, fire)
LOGGER.success('Data successfully inserted in db')
except Exception as error_msg:
LOGGER.error(f'Could not get currency data for {country} because of the following error: {error_msg}')
pass
result['operator'] = {'value': ''}
if 'phone' not in result:
result['phone'] = {'value': ''}
if 'email' not in result:
result['email'] = {'value': ''}
if 'website' not in result:
result['website'] = {'value': ''}
LOGGER.info(f'Getting embassy data for {result["country"]["value"]}')
DB.insert_or_update('embassies', result['country']['value'],
result['city']['value'],
result['operator']['value'],
result['type']['value'], result['phone']['value'],
result['email']['value'],
result['website']['value'])
LOGGER.success(
f'Successfully entered embassy information for {result["country"]["value"]}'
)
except Exception as error_msg:
LOGGER.error(
f'Could not get embassy data for {result["country"]["value"]} because of the following error: {error_msg}'
)
pass
for result in consulates_results["results"]["bindings"]:
try:
if 'city' not in result:
result['city'] = {'value': ''}
if 'operator' not in result:
result['operator'] = {'value': ''}
if 'phone' not in result:
result['phone'] = {'value': ''}
# Parse currencies and add to database
for country in iso_list:
try:
LOGGER.info(f'Beginning currency parsing for country: {country}')
url_converter = UrlHelper(currency_api_link)
information_link = url_converter.get_currency_api(country)
LOGGER.info(
f'Retrieving information from following link: {information_link}')
currency_api = ApiHelper(url_converter.get_currency_api(country))
LOGGER.info(f'Parsing returned HTML code: {currency_api.get_code()}')
data = currency_api.get_json()['currencies'][0]
LOGGER.success(f'Following currency data was retrieved: {data}')
for k, v in data.items():
if v is None:
data[k] = "None"
LOGGER.info('Inserting data into database.')
DB.insert_or_update('currencies', country, data['name'], data['code'],
data['symbol'])
except Exception as error_msg:
LOGGER.error(
f'Could not get currency data for {country} because of the following error: {error_msg}'
)
pass
